Mailman 3 July 2013 - scikit-image

Getting chain code of the shape in image
by Valeriy Sokolov July 15, 2013

July 15, 2013

Hi, I need to get a chain code of the region in the image. I failed to find from the docs any direct way to do it. Could you please suggest me the way how to do it with skimage? Or should I implement a boundary traverse by myself? Thank you in advance.

1 0

Building the docs using python 3
by jostein.floystad＠gmail.com July 14, 2013

July 14, 2013

Hi all, I've recently tried to make the transition to python3, at least for skimage development and testing. There is one issue stopping me: The documentation of skimage can't be built using python3 (keep in mind that we should also use a version of sphinx based on python3). Many of the issues that pop up when using python3 seem to be related to various sphinx extensions, for example scikit-image/doc/ext/numpydoc.py and scikit-image/doc/tools/{apigen.py,build_modref_templates.py}. I've seen several of these files have already been ported to python3 by other projects (numpy and ipython, at least), but I have not searched extensively for these newer versions. I assume that we do want the docs to build using python3. The question is how we go about porting the files in question with a minimal duplication of the efforts made by other projects? Just copying files into the skimage source tree seems very much like a short-term solution to me... On a related note, it would be nice to have travis CI build the docs and look for errors automatically. Presently, if an example fails to run, that example seems to simply disappear from the generated docs without the build failing; this is hardly ideal. Cheers! Jostein

3 2

Re: My porblem understanding Local Binary Pattern
by ginger wend July 14, 2013

July 14, 2013

Does it work with also with scikit-image 0.8 or only with 0.9dev? Im using 0.8 and got some error regrarding label2rgb. On Friday, July 12, 2013 6:30:06 PM UTC+2, Tony S Yu wrote: > > > > > On Fri, Jul 12, 2013 at 7:13 AM, ginger wend <gin...(a)walla.com<javascript:> > > wrote: > >> Hi, >> >> I am trying to learn Local Binary Pattern with scikit-image and as I have >> only limited knowledge in numpy and image analysis I find it hard. As a >> result, I will appreciate a lot if you can provide me some answers as I am >> sure that they will help me to advance: >> >> - If I understand correctly the >> skimage.feature.local_binary_pattern(image, P, R, method=' uniform) will go >> pixel by pixel (starting from pixel location 1,2) and generate the binary >> number based on the P and R, am I right?. If so what happens if the R is >> “out of boundary”? >> - How do you calculate the histogram and why are there are 18 bins in the >> example ( >> http://scikit-image.org/docs/dev/auto_examples/plot_local_binary_pattern.ht… >> )? >> - The output is (N, M) array and when I look at it is not clear to me >> what is the output and how do you generate a histogram from it, can someone >> you please clarify it. >> >> I know that these questions are be basic and while I was trying to look >> for answers I found the information more mathematical and less with applied >> approach. Therefore, any help, ideas or tips will help. >> >> Thanks for the patience, >> G. >> >> Hi, > > LBP can be quite confusing at first, but the idea is pretty simple. I > recently added some more explanations to the example in the gallery (most > of the addition is code to plot things---you don't really need to > understand that part of the code): > > > https://github.com/scikit-image/scikit-image/blob/master/doc/examples/plot_… > > That example should clarify most of your questions. I don't really know > how the boundary is treated, so maybe someone else can clarify that point. > > Best, > -Tony > >

3 3

Re: Qhull error
by Chintak Sheth July 14, 2013

July 14, 2013

> No, the number of pixels in the convex hull will be greater than or equal to the number of pixels in the object. Hence 1.0 is the upper bound. When using flag values, you usually want values outside the normal range, or at the edge of it at least! > Yes, rightly pointed out. > > Juan are you taking this issue up? > > I can look at it, but I have a busy few weeks ahead... We'll see which of us finishes their current agenda first! =) > Sounds good! :) Chintak

1 0

Re: Qhull error
by Juan Nunez-Iglesias July 14, 2013

July 14, 2013

@Chintak, > Juan, 0.9 might have been a more plausible value, since if I'm not mistaken it is "number of pixels in image / number of pixels in the convex hull". Convex hull image will have more number of pixels. ;) No, the number of pixels in the convex hull will be greater than *or equal to *the number of pixels in the object. Hence 1.0 is the upper bound. When using flag values, you usually want values outside the normal range, or at the edge of it at least! > Juan are you taking this issue up? I can look at it, but I have a busy few weeks ahead... We'll see which of us finishes their current agenda first! =) On Thu, Jul 11, 2013 at 8:04 PM, Chintak Sheth <chintaksheth(a)gmail.com>wrote: > Hi, > > Juan are you taking this issue up? Else I wouldn't mind taking a go at it > though I first need to complete the Cython implementation of Fast Marching > Method code. Once I have that all ready for the PR I don't mind taking a > shot here. > > Chintak > On Jul 11, 2013 4:12 PM, "Stéfan van der Walt" <stefan(a)sun.ac.za> wrote: > >> On Thu, Jul 11, 2013 at 12:32 PM, Chintak Sheth <chintaksheth(a)gmail.com> >> wrote: >> > First, I think we need to find why the dual nature, `convex_hull_image` >> > independently works fine but not from `regionprops`. >> >> Spot on. Thanks for following through! >> >> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscribe(a)googlegroups.com. >> For more options, visit https://groups.google.com/groups/opt_out. >> >> >> -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Re: seeking advice on HoG applicability
by Lisa Torrey July 13, 2013

July 13, 2013

Thanks! I'll look into alternative ways of producing features. -Lisa On Monday, July 8, 2013 6:28:52 AM UTC-4, Stefan van der Walt wrote: > > Hi Lisa > > Interestingly, Adam Wisniewski was working on this one-class > classification problem at the recent SciPy2013 sprint. Olivier Grisel > and Nelle Varoquaux from the sklearn team were able to give us some > helpful advice, and it might be worth getting in touch with them as > well. > > On Wed, Jul 3, 2013 at 9:10 PM, Lisa Torrey <lisa....(a)gmail.com<javascript:>> > wrote: > > - I have much less data. (Just 77 positives and 78 negatives, compared > to > > Dalal's 1239 and 12180.) > > You'll probably have to do some kind of cross-validation. > > > - My images aren't all the same size, like the pedestrian images are. > (I'm > > not sure if this would matter?) > > Perhaps investigate multi-scale texture features, such as the wavelet > coefficients (see http://www.pybytes.com/pywavelets/ ; even simple > statistics might suffice). > > > - My images are much higher resolution. (I've been downscaling them by a > > factor of 8, but the feature vectors are still enormous.) > > You'd want to extract some features that help the classifier, e.g. > daisy (http://scikit-image.org/docs/dev/auto_examples/plot_daisy.html), > texture features via grey-level co-occurrence matrices, or haralick > features (we don't yet have those in skimage, although they are > available in Luis Coelho's Mahotas). > > Regards > Stéfan >

2 3

Re: My porblem understanding Local Binary Pattern
by Johannes Schönberger July 12, 2013

July 12, 2013

The image is padded with 0s. Johannes Schönberger Am 12.07.2013 um 18:30 schrieb Tony Yu <tsyu80(a)gmail.com>: > > > > On Fri, Jul 12, 2013 at 7:13 AM, ginger wend <gingerw(a)walla.com> wrote: > Hi, > > I am trying to learn Local Binary Pattern with scikit-image and as I have only limited knowledge in numpy and image analysis I find it hard. As a result, I will appreciate a lot if you can provide me some answers as I am sure that they will help me to advance: > > - If I understand correctly the skimage.feature.local_binary_pattern(image, P, R, method=' uniform) will go pixel by pixel (starting from pixel location 1,2) and generate the binary number based on the P and R, am I right?. If so what happens if the R is “out of boundary”? > - How do you calculate the histogram and why are there are 18 bins in the example (http://scikit-image.org/docs/dev/auto_examples/plot_local_binary_pattern.ht… > - The output is (N, M) array and when I look at it is not clear to me what is the output and how do you generate a histogram from it, can someone you please clarify it. > > I know that these questions are be basic and while I was trying to look for answers I found the information more mathematical and less with applied approach. Therefore, any help, ideas or tips will help. > > Thanks for the patience, > G. > > Hi, > > LBP can be quite confusing at first, but the idea is pretty simple. I recently added some more explanations to the example in the gallery (most of the addition is code to plot things---you don't really need to understand that part of the code): > > https://github.com/scikit-image/scikit-image/blob/master/doc/examples/plot_… > > That example should clarify most of your questions. I don't really know how the boundary is treated, so maybe someone else can clarify that point. > > Best, > -Tony > > > -- > You received this message because you are subscribed to the Google Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > >

1 0

My porblem understanding Local Binary Pattern
by ginger wend July 12, 2013

July 12, 2013

Hi, I am trying to learn Local Binary Pattern with scikit-image and as I have only limited knowledge in numpy and image analysis I find it hard. As a result, I will appreciate a lot if you can provide me some answers as I am sure that they will help me to advance: - If I understand correctly the skimage.feature.local_binary_pattern(image, P, R, method=' uniform) will go pixel by pixel (starting from pixel location 1,2) and generate the binary number based on the P and R, am I right?. If so what happens if the R is “out of boundary”? - How do you calculate the histogram and why are there are 18 bins in the example (http://scikit-image.org/docs/dev/auto_examples/plot_local_binary_pattern.ht… - The output is (N, M) array and when I look at it is not clear to me what is the output and how do you generate a histogram from it, can someone you please clarify it. I know that these questions are be basic and while I was trying to look for answers I found the information more mathematical and less with applied approach. Therefore, any help, ideas or tips will help. Thanks for the patience, G.

2 1

Re: Qhull error
by Chintak Sheth July 11, 2013

July 11, 2013

Hi, Juan are you taking this issue up? Else I wouldn't mind taking a go at it though I first need to complete the Cython implementation of Fast Marching Method code. Once I have that all ready for the PR I don't mind taking a shot here. Chintak On Jul 11, 2013 4:12 PM, "Stéfan van der Walt" <stefan(a)sun.ac.za> wrote: > On Thu, Jul 11, 2013 at 12:32 PM, Chintak Sheth <chintaksheth(a)gmail.com> > wrote: > > First, I think we need to find why the dual nature, `convex_hull_image` > > independently works fine but not from `regionprops`. > > Spot on. Thanks for following through! > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Re: Error in plot_matching.py example
by Johannes Schönberger July 11, 2013

July 11, 2013

Strange… have you tried to do a completely clean new install (new virtualenv, new clone of git repo etc.)? Johannes Schönberger Am 11.07.2013 um 19:44 schrieb Ankit Agrawal <aaaagrawal(a)gmail.com>: > Hi Johannes, > > On Fri, Jul 12, 2013 at 1:24 AM, Johannes Schönberger <jschoenberger(a)demuc.de> wrote: > I guess you are in the root directory of the scikit-image git repository and hence the skimage folder is in your namespace. Either: > > cd doc/examples; python plot_matching.py > > or: > > make; python doc/examples/plot_matching.py > > Both of the above ways return the same error. > > Cheers, > Ankit. > > -- > You received this message because you are subscribed to the Google Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > >

1 0